day711_28imageclassification mnist.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from keras.datasets import mnist\n",
    "import random\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "(xtr,ytr),(xts,yts) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_true=yts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "xtr shape:(60000, 28, 28)\n",
      "ytr shape:(60000,)\n"
     ]
    }
   ],
   "source": [
    "print('xtr shape',xtr.shape,sep=':')\n",
    "print('ytr shape',ytr.shape,sep=':')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 50 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "n_cols = 5\n",
    "n_classes = 10\n",
    "fig,axes = plt.subplots(nrows=n_classes,ncols=n_cols,figsize=(15,15))\n",
    "fig.tight_layout()\n",
    "for i in range(n_cols):\n",
    "    for j in range(n_classes):\n",
    "        images = xtr[ytr==j]\n",
    "        selected_image = images[random.randint(0,(len(images)-1)),:,:]\n",
    "        axes[j][i].imshow(selected_image,cmap='gray')\n",
    "        axes[j][i].axis('off')\n",
    "        if i==2:\n",
    "            axes[j][i].set_title(str(j))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.countplot(ytr)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Pre-processing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_pixels = xtr.shape[1]*xtr.shape[2]\n",
    "xtr = xtr.reshape(xtr.shape[0],n_pixels)\n",
    "xtr = xtr/255\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "xts = xts.reshape(xts.shape[0],n_pixels)\n",
    "xts = xts/255"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# OneHot Encoding ytr array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.utils.np_utils import to_categorical\n",
    "ytr=to_categorical(ytr,10) #10 represents number of classes/categories \n",
    "yts= to_categorical(yts,10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(60000, 10)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ytr.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Creating neural net model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.optimizers import Adam"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 32)                25120     \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 64)                2112      \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 10)                650       \n",
      "=================================================================\n",
      "Total params: 27,882\n",
      "Trainable params: 27,882\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "def create_model():\n",
    "    model = Sequential()\n",
    "    model.add(Dense(32,input_dim=n_pixels,activation='relu'))\n",
    "    model.add(Dense(64,activation='relu'))\n",
    "    model.add(Dense(10,activation='softmax'))  #outputnode for images in case of images activation will always be softmax\n",
    "    model.compile(Adam(lr=0.01),loss='categorical_crossentropy',metrics=['accuracy'])\n",
    "    return model\n",
    "model = create_model()\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 60000 samples, validate on 10000 samples\n",
      "Epoch 1/10\n",
      "60000/60000 [==============================] - 3s 46us/step - loss: 0.2797 - accuracy: 0.9166 - val_loss: 0.1613 - val_accuracy: 0.9529\n",
      "Epoch 2/10\n",
      "60000/60000 [==============================] - 3s 43us/step - loss: 0.1803 - accuracy: 0.9479 - val_loss: 0.1907 - val_accuracy: 0.9480\n",
      "Epoch 3/10\n",
      "60000/60000 [==============================] - 3s 42us/step - loss: 0.1630 - accuracy: 0.9528 - val_loss: 0.1803 - val_accuracy: 0.9519\n",
      "Epoch 4/10\n",
      "60000/60000 [==============================] - 3s 49us/step - loss: 0.1503 - accuracy: 0.9570 - val_loss: 0.1772 - val_accuracy: 0.9516\n",
      "Epoch 5/10\n",
      "60000/60000 [==============================] - 3s 43us/step - loss: 0.1420 - accuracy: 0.9598 - val_loss: 0.1608 - val_accuracy: 0.9587\n",
      "Epoch 6/10\n",
      "60000/60000 [==============================] - 3s 44us/step - loss: 0.1328 - accuracy: 0.9619 - val_loss: 0.1647 - val_accuracy: 0.9593\n",
      "Epoch 7/10\n",
      "60000/60000 [==============================] - 3s 43us/step - loss: 0.1264 - accuracy: 0.9641 - val_loss: 0.1527 - val_accuracy: 0.9587\n",
      "Epoch 8/10\n",
      "60000/60000 [==============================] - 3s 46us/step - loss: 0.1249 - accuracy: 0.9659 - val_loss: 0.1456 - val_accuracy: 0.9654\n",
      "Epoch 9/10\n",
      "60000/60000 [==============================] - 3s 47us/step - loss: 0.1180 - accuracy: 0.9681 - val_loss: 0.1671 - val_accuracy: 0.9613\n",
      "Epoch 10/10\n",
      "60000/60000 [==============================] - 3s 47us/step - loss: 0.1151 - accuracy: 0.9692 - val_loss: 0.1991 - val_accuracy: 0.9556\n"
     ]
    }
   ],
   "source": [
    "h = model.fit(xtr,ytr,epochs=10,verbose=1,validation_data=(xts,yts),batch_size=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save('model.h5')   #saving the model with good accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predicting images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "import cv2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "img=cv2.imread(r\"C:\\Users\\amanv\\Documents\\ai&mlclass\\Datasets\\digit.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(389, 385, 3)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "img.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img=cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)\n",
    "img=cv2.resize(img,(28,28))\n",
    "plt.imshow(img,cmap='gray')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "img = img.reshape(1,n_pixels)\n",
    "img = img/255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([6], dtype=int64)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.predict_classes(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 964    0    5    0    0    3    4    0    2    2]\n",
      " [   0 1102    5    4    2    1    3    0   12    6]\n",
      " [   3    1 1012    2    2    0    1    5    6    0]\n",
      " [   1    0   23  957    0   14    0    5    3    7]\n",
      " [   3    1    2    1  912    0    9    2    1   51]\n",
      " [   4    0    1   11    0  862    7    1    1    5]\n",
      " [   7    3    1    1    5    7  928    0    3    3]\n",
      " [   1    9   17    4    0    0    1  933    2   61]\n",
      " [  13    0   13    6    2   19    0   10  900   11]\n",
      " [   2    1    1    4    4    6    0    4    1  986]]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "y_pred = model.predict_classes(xts)\n",
    "print(confusion_matrix(y_true,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}